[1]张瀚元,赵博伟,胡 伦*,等.基于图注意力网络的环状 RNA 与疾病关联关系预测[J].计算机技术与发展,2023,33(11):126-134.[doi:10. 3969 / j. issn. 1673-629X. 2023. 11. 019]
 ZHANG Han-yuan,ZHAO Bo-wei,HU Lun*,et al.Prediction of Circ RNA-Disease Associations Based on Graph Attention Networks[J].,2023,33(11):126-134.[doi:10. 3969 / j. issn. 1673-629X. 2023. 11. 019]
点击复制

基于图注意力网络的环状 RNA 与疾病关联关系预测()
分享到:

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
33
期数:
2023年11期
页码:
126-134
栏目:
人工智能
出版日期:
2023-11-10

文章信息/Info

Title:
Prediction of Circ RNA-Disease Associations Based on Graph Attention Networks
文章编号:
1673-629X(2023)11-0126-09
作者:
张瀚元1 赵博伟1 胡 伦1* 王 磊2* 尤著宏3
1. 中国科学院大学 中国科学院新疆理化技术研究所,新疆 乌鲁木齐 830011;
2. 广西科学院 大数据与智能计算研究中心,广西 南宁 530007;
3. 西北工业大学 计算机学院 大数据存储与管理工业和信息化部重点实验室,陕西 西安 710072
Author(s):
ZHANG Han-yuan1 ZHAO Bo-wei1 HU Lun1* WANG Lei2* YOU Zhu-hong3
1. Xinjiang Technical Institute of Physics and Chemistry,Chinese Academy of Sciences,Urumqi 830011,China;
2. Big Data and Intelligent Computing Research Center,Guangxi Academy of Sciences,Nanning 530007,China;
3. MIIT Key Laboratory of Big Data Storage and Management,School of Computer Science,Northernwestern Polytechnic University,Xi’n 710072,China
关键词:
环状 RNA / CircRNA疾病关联关系预测图注意力网络深度学习
Keywords:
Circular RNA / CircRNAdiseaseassociation predictiongraph attention networksdeep learning
分类号:
TP399
DOI:
10. 3969 / j. issn. 1673-629X. 2023. 11. 019
摘要:
环状 RNA 是一种具有环状结构并且表达水平与多种疾病有关的非编码 RNA 分子,挖掘环状 RNA 与疾病之间的内在关联关系在生命医学研究中具有重要意义。 基于图注意力机制,
该文提出了一种由图注意力网络( GAT) 、编码器-解码器(AE) 和全连接神经网络( DNN) 结构组合的端到端深度学习模型 GATECDA 来预测潜在的环状 RNA 与疾病的关联关系。?
在包含 739 个关系的 CircR2Disease 数据集上,GATECDA 模型五折交叉验证实验取得了 ROC 曲线下面积 AUC 为0郾 961 8,AUPR 为 0. 903 2,衡量在非平衡数据上性能 MCC 指标达到了 0. 757 6 的优异结果,综合性能在同领域预测模型中表现出色。 表明基于深度学习图表示学习的策略有助于提升环状 RNA 与疾病关联关系预测模型的综合性能,同时端到端的学习模型更易于训练与泛化到其他问题中。 在预测的结果得到的前 30 个环状 RNA 与疾病的关联关系中,有 25 个在最近医学文献中有支持。 表明人工智能方法可以为医学研究筛选与疾病相关的标志物提供新的角度。
Abstract:
Circular RNA ( CircRNA) is a kind of expressed RNA transcript with loop structure and its expressed level related to other diseases. It is of great significance to explore the internal correlation between CircRNA and Disease in life medicine research. Based on thegraph attention mechanism,GATECDA,an end-to-end deep learning model consisting?
of graph attention network ( GAT) ,AutoEncoder( AE) and deep neural network ( DNN) ,is proposed to predict the candidate associations between CircRNA and Disease. It achieved 5-fold cross-validation on AUC at 0. 961 8 and AUPR at 0. 903 2, MCC index at 0. 757 6 on CircR2Disease data set including 739associations between CircRNA?
and Disease. The measurement result means the model performed well on the imbalanced benchmark.Hereby,we believed the strategy by integrating graph attention network embedding into the deep learning model would improve the performance of prediction CircRNA-Disease association. At top 30 of the predicted association of CircRNA and Disease,we retrieved 25 ofthem with published paper supporting. As we thought that the AI tech. would boost the work of discovering biomarkers related with disease.
更新日期/Last Update: 2023-11-10